Twelve types of electronic identification ruminal boluses of different dimensions were used to obtain a model for predicting their retention in the forestomachs of sheep. Boluses (n = 1,662) were made of ceramic materials, and their dimensions varied in o.d. (9 to 21 mm), length (37 to 68 mm), volume (2.5 to 21.0 mL), and specific gravity (0.85 to 3.91). Each bolus contained a half-duplex, standardized, glass-encapsulated transponder (32 × 3.8 mm). Boluses were administered to sheep (n = 1,497) of different ages by using the appropriate balling guns, and their retention under semiintensive conditions was recorded for at least 2 yr. When a bolus was lost, the sheep was rebolused with a heavier bolus. All sheep wore 2 plastic ear tags: one for the official control of health programs and the other for farm use. To determine the anatomical limit for a bolus passing through the gastrointestinal tract, the size of the reticulo-omasal orifice was measured in 46 adult sheep (male, n = 14; female, n = 32) that died by causes not related to bolus administration during the experiment. No signs of disease or growth alteration were detected in the bolused sheep. Total ear tag losses during the experiment period were 7.5% on average. Bolus retention (5 to 100%) varied according to bolus features and age of the sheep, but it showed a plateau after 18 mo. Inadequately dimensioned boluses were regurgitated or passed through the gastrointestinal tract and were excreted with the feces. The diameter of the reticulo-omasal orifice in adult sheep differed between male and female (23.1 and 21.8 mm, respectively; P < 0.01) and was greater than the o.d. of the retained boluses. Retention rate was predicted from bolus weight and volume by a logistic regression (R2 = 0.997; P < 0.001). When retention data from the literature (59.0 to 100%) were included in the model, the adjustment was slightly lower (R2 = 0.967). As a result, the minimum bolus weight estimated to reach a 99.5% retention rate in sheep varied between 16 and 45 g when volume varied between 3 and 22 mL, for boluses with a specific gravity between 2.0 and 5.2. In conclusion, bolus retention rate in sheep varied dramatically according to their features. For safe and efficient retention of electronic identification boluses in sheep, boluses of small volume and diameter (e.g., <15 mm) with specific gravity and weight greater than 3.0 and 20 g, respectively, are recommended.